The current solution for the lock doesn't work properly and causes a short spike in user traffic to have too many clippy-service processes scheduled and run at the same time, causing 100% CPU usage and unresponsiveness of the service. While the important solution here is to have the worker systems of #15 up and running, we should think about fixing the lock either way – and probably first. Take a look at http://redis.io/topics/distlock#correct-implementation-with-a-single-instance for information on how to do that.
The current solution for the lock doesn't work properly and causes a short spike in user traffic to have too many clippy-service processes scheduled and run at the same time, causing 100% CPU usage and unresponsiveness of the service. While the important solution here is to have the worker systems of #15 up and running, we should think about fixing the lock either way – and probably first. Take a look at http://redis.io/topics/distlock#correct-implementation-with-a-single-instance for information on how to do that.